This closes #3808.

NOT patterns are not yet supported when an OPTIONAL
pattern directly preceeds a NOT. In these cases, an
exception is thrown that proposes an alternative
(but not the most efficient) way to support these
patterns.

This closes #3692.

This closes #3648.

The AsyncWaitOperator should not only react to orderly completed
StreamElementQueueEntries but also to those completed with a user exception
or those which timed out.

This PR fixes the problem by calling the onComplete function passed to
StreamElementQueueEntry#onComplete also in the exceptional case.

This closes #3814.

Remove client less factory methods from ZooKeeperUtils

Introduce default job id

This closes #3781.

This PR introduces a standalone high availability services implementation which can be used
in a distributed setting with no HA guarantees. Additionally, it introduces a common base
class which is also used by the EmbeddedHaServices. This base class instantiates the
standalone variants of the checkpoint recovery factory, submitted job graphs store, running
jobs registry and blob store.

The StandaloneHaServices are instantiated with a fixed address for the Job- and
ResourceManager. This address and the HighAvailability.DEFAULT_LEADER_ID is returned by
the corresponding LeaderRetrievalServices when being started.

This closes #3622.

This closes #3604

This closes #3809.

This closes #3741.

This closes #3769.

This closes #3709.

This closes #3780.

This check is present in WindowedStream but was missing on
AllWindowedStream.

This closes #3791.

  - Rename 'ExecutionGraph.fail()' to 'ExecutionGraph.failGlobally()' to differentiate from fine grained failures/recovery
  - Add base class for FailoverStrategy
  - Add default implementation (restart all tasks)
  - Add logic to load the failover strategy from the configuration

The updated curator version includes a bugfix for a potential block

The binary snapshots have been created on the Flink 1.2 branch.

The binary snapshots have been created on the Flink 1.2 branch.

The binary snapshots for this were created on the Flink 1.2 branch.

The binary snapshots were created on the Flink 1.2 branch.

The binary snapshots were created on the Flink 1.2 branch.

This has methods for storing/reading OperatorStateHandles, as returned
from stream operator test harnesses. This can be used to write binary
snapshots for use in state migration tests.

The binary savepoints in this were created on the Flink 1.2.0 release
commit.

In Flink 1.2 we only wrote timers to key-grouped state streams. With the
addition of the watermark callback service we started to also write the
watermark callbacks to the key-grouped streams. This breaks backwards
compatibility with saveoints taken on Flink 1.2

This replaces usage of the watermark callback service by setting a
regular timer for "current watermark + 1". These timers will fire as
soon as the watermark advances, thus simulating the watermark callback
service.

State that was checkpointed using Checkpointed (on a user function)
could be restored using CheckpointedRestoring when the savepoint was
done on Flink 1.2. The reason was an overzealous check in
AbstractUdfStreamOperator that only restores from "legacy" operator
state using CheckpointedRestoring when the stream is a Migration stream.

This removes that check but we still need to make sure to read away the
byte that indicates whether there is legacy state, which is written when
we're restoring from a Flink 1.1 savepoint.

After this fix, the procedure for a user to migrate a user function away
from the Checkpointed interface is this:

 - Perform savepoint with user function still implementing Checkpointed,
   shutdown job
 - Change user function to implement CheckpointedRestoring
 - Restore from previous savepoint, user function has to somehow move
   the state that is restored using CheckpointedRestoring to another
   type of state, .e.g operator state, using the OperatorStateStore.
 - Perform another savepoint, shutdown job
 - Remove CheckpointedRestoring interface from user function
 - Restore from the second savepoint
 - Done.

If the CheckpointedRestoring interface is not removed as prescribed in
the last steps then a future restore of a new savepoint will fail
because Flink will try to read legacy operator state that is not there
anymore.  The above steps also apply to Flink 1.3, when a user want's to
move away from the Checkpointed interface.